LT350 Wants to Put Data Centers on Parking Lot Canopies

A neurosurgeon-turned-inventor is pitching a new answer to the data center industry’s biggest question: where to put them. Jeff Thramann, CEO of LT350, wants to install small modular data centers in the airspace above open-air parking lots, according to a profile published by Commercial Observer this month.

Thramann, a Cornell Medical School graduate who lives in Boulder, Colo., has at least 130 patents to his name. His earlier startups include U.S. Radiosurgery, sold to a publicly traded company, and Denver CyberKnife, a cancer treatment facility bought by a private firm. He formed LT350 last year with former U.S. Secretary of Defense Mark Esper.

The concept is straightforward. LT350 builds solar-paneled parking canopies at sites such as malls, hospitals and schools. Beneath the solar roof, cartridge boxes hold GPU servers alongside battery storage and inverters. The company says each canopy has a footprint of roughly 2,000 square feet, while the GPUs can draw nearly 1 megawatt of electricity, enough to supply between 400 and 1,000 U.S. homes at any moment. No parking spaces are lost.

The battery system is central to the design. Thramann said the batteries charge during off-peak hours and, at peak times, the units run off the grid so the grid does not see the load. Before building, LT350 assesses a lot’s electrical capacity and the grid network it connects to, then sizes the batteries to avoid drawing too much power.

Thramann traced the idea to his long interest in solar and battery technology, which began about 20 years ago. He bought Tesla car No. 541 off the assembly line in 2008, but concluded that battery intermittency and costly commercial solar maintenance were persistent problems. He filed patents for battery cartridges housed in canopy ceilings, then layered on EV charging. Data centers, he said, became the missing piece, since he could use parking-lot airspace while also helping the grid.

The target is inference, the real-time work of running pre-trained AI models to answer user requests. Thramann argues that placing compute closer to users, in residential, commercial and retail settings, offers an edge over giant facilities built far from population centers. Health care is the lead market, because hospitals worry about protecting personal health information as data flows through AI models and the cloud. A rooftop unit in a hospital lot, he said, could run a cloud model behind the facility’s firewall. Financial services, government and defense contractors are also targets.

LT350 also sees hyperscalers, so-called neo clouds and frontier model labs as potential co-location customers. “They’re all trying to get closer,” Thramann said, “but their question is: Where can we put these things to be close? And we think we have that solution.”

The company is still early. LT350 is pre-revenue and still designing its models, though it forecasts $12.4 million in annual revenue per 2,000-square-foot unit. It is currently involved in a merger with Auddia, a publicly traded AI audio technology company Thramann founded with a $6 million market cap. The combined entity will be owned by Thramann Holdings, which is valued at $250 million.

On the ground, LT350 plans to break ground early next year on its first data center parking lot at a hospital in Dallas. It is negotiating with quick-service restaurants, convenience stores and gas stations, anything that already has canopies, as well as malls and big-box retail. For New York, where land is scarce, power is tight and surface lots are common at outer-borough retail and medical campuses, the model offers a notable thought experiment, though the company has announced no plans here. As with any pre-revenue venture, the forecasts remain unproven.